Development of a Graph Neural Network for Processing Text Data

This article examines graph-based modeling of text data using neural networks. The aim of this study is to develop a graph neural network for text classification and clustering based on semantic content. Texts are represented as graphs, where vertices are concepts and edges are connections between them. Public text corpora in Russian and English are used. A new approach to analyzing text data is proposed based on their representation as oriented weighted graphs and processing by graph neural networks. Graphs are processed by a neural network with three layers of graph convolutions. The obtained results show accuracy exceeding 90% for topic group classification and text clustering, outperforming RNN, CNN, and doc2vec methods.